REVIEW doi:10.

1038/nature14434

The deuterostome context

of chordate origins
Christopher J. Lowe1, D. Nathaniel Clarke1, Daniel M. Medeiros2, Daniel S. Rokhsar3,4,5 & John Gerhart3

Our understanding of vertebrate origins is powerfully informed by comparative morphology, embryology and genomics
of chordates, hemichordates and echinoderms, which together make up the deuterostome clade. Striking body-plan dif-
ferences among these phyla have historically hindered the identification of ancestral morphological features, but recent
progress in molecular genetics and embryology has revealed deep similarities in body-axis formation and organization
across deuterostomes, at stages before morphological differences develop. These developmental genetic features, along
with robust support of pharyngeal gill slits as a shared deuterostome character, provide the foundation for the emergence
of chordates.

T
he mystery of chordate origins has endured for more than
150 years. Shortly after Darwin’s On the Origin of Species, acorn
worms were discovered to have chordate-like pharyngeal gill
slits1,2 and to metamorphose from echinoderm-like larva3, thus linking
the evolution of chordates, hemichordates and echinoderms. Modern
phylogenetic analysis has confirmed the union of these three phyla in a
single clade. This group, the deuterostomes, provides the phylogenetic
framework for developing hypotheses about the origin of chordate fea-
tures through comparative morphology, embryology and genomics.
The emergence of comparative molecular developmental biology over
the past quarter of a century has revived interest in classic hypotheses
of animal body-plan evolution4. The comparative approach focuses
on identifying morphological, developmental and genetic traits that
are shared across phyla by virtue of their inheritance from a common
ancestor, and provides an understanding of how such ancestral traits
can arise and be subsequently modified. Although many recent hypoth-
eses on chordate and vertebrate origins on the basis of molecular data
are motivated primarily by projections from the bilaterian ancestor4,5, a
growing body of data from hemichordates, echinoderms and inverte-
brate chordates serves as the foundation for new hypotheses based on
deuterostome ancestral characters6–14.
Despite the impressive morphological disparity among deuterostome
phyla, we are making progress identifying conserved anatomical and
molecular ancestral characters. Each phylum is a fascinating natural
experiment in body-plan evolution, but their dazzling diversity presents
a major challenge for reconstructing early deuterostome evolutionary
history in morphological terms (Box 1)15. In this Review we highlight
recent advances in deuterostome phylogenetics, developmental biology
and genomics that have contributed to our understanding of the early
evolution of deuterostomes and the subsequent origin of chordates.
Deuterostome phylogeny
The first step in unravelling chordate origins is the establishment of a
robust deuterostome phylogeny (Fig. 1). The chordates, uniting verte-
brates, tunicates and cephalochordates, were first recognized by Hae-
ckel16, partly based on shared developmental characteristics. A key insight
came from Kowalevsky’s17 recognition that the tadpole larva of ascidians
shared many characteristics with vertebrates, an observation that greatly
1
Hopkins Marine Station, Department of Biology, Stanford University, 120 Oceanview Boulevard, Pacific Grove, California 93950, USA. 2Department of Ecology and Evolutionary Biology, 
University
of Colorado, Boulder
Ramaley N122 Campus Box 334
Boulder, Colorado 80309-0334, USA. 3Department of Molecular and Cell Biology
University of California, Berkeley
142 Life Sciences Addition
3200
Berkeley, California 94720-3200, USA. 4Okinawa Institute of Science and Technology 1919-1 Tancha, 
Onna, Okinawa Prefecture 904-0495, Japan. 5Department of Energy Joint Genome
Institute 2800 Mitchell Drive
Walnut Creek, California 94598, USA.
impressed Darwin18. Kowalevsky also recognized the vertebrate-like gill
slits of the invertebrate acorn worms2. The link between chordates and
acorn worms was emphasized by Bateson, who proposed further mor-
phological affinities between them in the late 1800s, and named the acorn
worms ‘hemichordates’1. Around the same time, Metchnikoff recognized
the similar larval forms of hemichordates and echinoderms, and united
these two phyla into the ‘Ambulacraria’3 (Box 2).
The unity of chordates, hemichordates and echinoderms was inferred
by Grobben19 on the basis of three shared developmental features: ‘deu-
terostomous’ development (derivation of the mouth from a secondary
opening rather than the blastopore), radial cleavage and enterocoely (the
pouching out of mesoderm from the archenteron wall). Although he
named this lineage the ‘deuterostomes’ (second mouth), we now recognize
that these features are not unique to the chordate–hemichordate–echi-
noderm clade, and are found in several other phyla20, the result of either
shared ancestry or convergence. This leads to the nomenclatural embar-
rassment that some phyla with deuterostomous development are not deu-
terostomes. Nevertheless the name has stuck, and by convention we refer
to the chordate–hemichordate–echinoderm clade as the deuterostomes.
The advent of molecular phylogenomics has brought new methods
to bear on the relationships between and within deuterostome phyla
(Fig. 1). Ambulacraria, the surprising grouping of hemichordates and
echinoderms, is strongly supported by molecular characters15,21–23, and
is clearly the sister group of chordates. Within chordates, it is now widely
recognized that the cephalochordate lineage (amphioxus) diverged
before the split between tunicates and vertebrates21,24. This recent dis-
covery overturned earlier thinking that tunicates diverged first, which
had implied that the simple ascidian tadpole larva represents ancestral
chordate features (Box 1).
Although classic embryological criteria suggested that lophophorates
(phoronids, brachiopods and bryozoans) and/or chaetognaths should also
be grouped among the deuterostomes, molecular phylogenetics robustly
supports their position in the protostomes22,25,26. Xenoturbellid worms are
a more challenging case: these animals resemble acoelomorphs (acoel
flatworms and nematodermatids) and have been grouped with them in
a ‘Xenacoelomorpha’ clade27,28. Some molecular analyses also identify
Xenoturbella and its relatives as ambulacrarians, and therefore deuter-
ostomes27, whereas other studies find that acoelomorphs diverge from

4 5 6 | NAT U R E | VO L 5 2 0 | 2 3 A P R I L 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved
 REVIEW INSIGHT

BOX 1

Deuterostome diversity
Unambiguous homologies between deuterostome phyla with a Chordates
morphologically disparate body plans are difficult to establish, leading Tunicates Vertebrates

to a wide range of often contradictory hypotheses about chordate

origins1,6,29,96,100,107,108,114–116. We present basic descriptions of the
adult body plans of the uncontested deuterostome phyla: chordates,
hemichordates and echinoderms. All mesodermally derived-structures Cephalochordates
are red, ectoderm are blue and endoderm are yellow. a, Chordates are
set apart from other deuterostomes by a suite of features that enable Nerve cord Nerve
swimming by paired muscles along a trunk that extends post-anally. ganglion
Notochord
These muscles exert forces on the notochord, a flexible rod that provides
elastic recoil to power movement. Chordates also have a unique Notochord Nerve
cord Anus
tubular central nervous system (CNS)29. Of the subphyla, vertebrates Anus
Branchial
are distinguished from other chordates by the elaboration of the basket
head region with an enlarged anterior CNS with paired sense organs,
evident here in a lamprey ammocoete larva (top) and an axolotl tadpole Mouth
(bottom). Tunicates (larvaceans, ascidians and thalacians) are a diverse
group of marine filter feeders that display a range of body plans and life- b Hemichordates
Pterobranchs Trunk
history strategies, including solitary, colonial, sessile and free-swimming Stomochord r
lla
forms117. They are represented here by ascidians. Chordate affinities are Co Gill slits
Enteropneusts Gill pore

cis
most evident in the larval form: an ascidian tadpole (left) has a tubular Pharynx
Harrimaniids

bos
Torquaratorids
Spengelids
Ptychoterids
Pterobranchs
nerve cord, a notochord and a post-anal tail. These features regress at Gonad

Pro
metamorphosis, leaving the branchial basket, a small nerve ganglion
and the endostyle as the only chordate characters remaining in the Enteropneusts
Heart–kidney
adult (right). Cephalochordates, represented by amphioxus, are filter complex
feeders that burrow in sand with their mouths open to the water column. Mouth
Amphioxus shares much of its basic anatomy with vertebrates, including
segmented musculature, and a vertebrate-like heart and circulatory Hepatic caeca
system30,118. They have a modest CNS consisting of a neural tube with
simplified vertebrate-type patterning along both the anteroposterior c Echinoderms
and dorsoventral axes8,118,119. b, Hemichordates are a clade of marine Crinoids Asteroids Ophiuroids Holothuroids Echinoids
worms divided into two groups: enteropneusts and pterobranchs.
Hemichordate phylogeny is based on Cannon et al.120. Pterobranchs,
shown here by Cephalodiscus, are small largely colonial animals that
live within the protection of a secreted fibrous tube and use a ciliated
lophophore for filter feeding50,58. Enteropneusts, or acorn worms, are
solitary, burrowing worms that feed using a combination of deposit and
Madreporite
filter feeding52,121. The harrimaniid Saccoglossus kowalevskii, which has
Anus
been used for many developmental studies12, is pictured (micrograph).
Both groups of hemichordates are united by their tripartite body plan, Stomach
Digestive glands Gonad
which includes proboscis, collar and trunk (as shown in the illustration of
a spengelid entropneust). The proboscis is used for digging and feeding
and contains the gut diverticulum called the stomochord that supports
Mouth
a heart–kidney complex56,60. The mouth opens ventrally into the pharynx
Radial canal
within the collar region, and the anterior trunk is perforated by a series Radial nerve Ring
of dorsolateral gill slits58. c, Echinoderms have considerably modified canal
Tube feet
the ancestral bilaterian body plan to become pentaradially symmetrical Nerve
ring
as adults, although their larvae are bilaterally symmetric (Box 2). Even
basic axis comparisons with other deuterostomes are problematic, mesodermally derived water vascular system, a hydraulic system that
and the evolutionary origins of this phylum remain a mystery. All five drives the distinctive tube feet used for feeding and locomotion; five
extant classes of echinoderms: crinoids (sea lilies), asteroids (sea stars), radial nerves along each arm/ambulacrum linked by a nerve ring, and
ophiuroids (brittle stars), holothuroids (sea cucumbers) and echinoids the mesodermally derived skeleton. Asteroids most clearly exhibit the
(sea urchins) are characterized by a conserved body plan shown by a basic components of the body plan. Phylogenetic relationships are
diagram of an asteroid with cutaways to show internal anatomy; the based on refs 120, 122.

the bilaterian stem before the protostome–deuterostome split28 (Fig. 1).
We note, however, that even if xenoturbellids and/or acoelomorphs are
deuterostomes, their simple body plans would represent secondary loss
from a more complex deuterostome ancestor. The resolution of the phylo-
genetic placement of these taxa is therefore unlikely to provide substantial
insight into vertebrate origins.
Ancestral chordate characters
On the basis of shared features of living chordates we have gained a rather
detailed view of the development, morphology and life history of the
last common chordate ancestor. Most classic and modern reconstruc-
tions of ancestral chordates propose a filter feeder with a notochord, gill
slits, endostyle, dorsal hollow nerve cord and post-anal tail29. The recent

2 3 A P R I L 2 0 1 5 | VO L 5 2 0 | NAT U R E | 4 5 7
© 2015 Macmillan Publishers Limited. All rights reserved
 INSIGHT REVIEW
revision of the chordate family tree has added to this list of ancestral chor-
date features. The basal position of cephalochordates among chordates
suggests that similarities between amphioxus and vertebrates represent
ancestral chordate features lost in tunicates21,24. Thus, in addition to the
core features listed earlier, the Early Cambrian or Pre-Cambrian chor-
date ancestor probably possessed myomeres, a vertebrate-like circulatory
system and a central nervous system (CNS)30. The life history of cephalo-
chordates, and the fact that larval lampreys and adult hemichordates are
burrowing filter feeders, further suggest that this ancestor was a solitary,
endobenthic filter feeder that was capable of short swims.
The striking similarities between amphioxus- and vertebrate-devel-
opmental mechanisms allow a fairly comprehensive reconstruction of
early development in primitive chordates. As in vertebrates and cepha-
lochordates, the anteroposterior (AP) and dorsoventral (DV) axes of the
ancestral chordate were probably determined during gastrula stages by
organizing centres much like Spemann’s organizer of vertebrates, secret-
ing long-range patterning signals11. Opposing Nodal and BMP signalling
gradients established the DV axis, with Chordin-mediated BMP inhibi-
tion in the dorsal ectoderm segregating the presumptive CNS from the
epidermal (or general) ectoderm11,31. Along the AP axis, Wnt and retinoic
acid signalling probably acted on Hox genes and other transcription-fac-
tor genes to establish the regional identities of AP domains of the body
axis, including the boundary between the foregut and hindgut and the
main subdivisions of the CNS8,32,33.
Comparisons between amphioxus and vertebrates suggest a deep
ancestry of the major divisions of the CNS along the AP axis. Later in
development, fine-scale patterning of the ancestral chordate CNS was also
vertebrate-like, but simpler. Along the DV axis of the CNS, all chordates
have a molecularly distinct dorsal domain that expresses pax3/7, msx and
zic genes and generates sensory interneuron cells34, a ventral floor plate
expressing hedgehog ligands35, and an intervening bilateral domain flank-
ing the neural tube lumen and generating motor and visceral neurons.
The expression domains of transcription factors and signalling molecules
along the AP axis of the CNS are also mostly conserved across chordates,
and presumably reflect expression domains of the chordate ancestor8.
Precisely how this patterning was generated is less clear, as current data
suggest that neither amphioxus nor tunicates have unambiguous, func-
tionally validated homologues of two vertebrate CNS signalling centres,
the isthmic organizer or the zona limitans intrathalamica (although these
signalling mechanisms may have been present in a deuterostome ances-
tor, see later)8,36,37.
Segmented musculature of the ancestral chordate almost certainly
developed from somites, and at least some formed by enterocoely35,38. In
amphioxus, the anterior-most somites form by enterocoely, whereas pos-
terior somites pinch off sequentially from the tail bud36,39. In vertebrates,
a ‘clock and wavefront’ mechanism, involving oscillating Notch and Wnt
Bilaterians
Deuterostomes
Chordates
Cnidarians Protostomes Vertebrates Tunicates Cephalochordates
Figure 1 | Deuterostome phylogeny.  A consensus cladogram of
deuterostome groups based on recent phylogenomic data sets21,22,24,28,113. There
are three major phyla of extant deuterostomes, which are grouped into two
diverse clades: the ambulacrarian phyla (green), consisting of hemichordates
4 5 8 | NAT U R E | VO L 5 2 0 | 2 3 A P R I L 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved
signalling and a posterior fibroblast growth factor (FGF)-signalling gradi-
ent divides the paraxial mesoderm into a series of somites40. Despite these
mechanistic differences, amphioxus displays vertebrate-like segmental
expression of Notch and Wnt signalling components in nascent somites,
and requires FGF signalling for forming and maturing the anterior and
posterior somites41,42. Thus, somitogenesis in all living chordates, and pre-
sumably their last common ancestor, involved iterated Notch–Delta and
Wnt signalling, and FGFs.
Despite differences in when and how the pharyngeal gill slits form in
the three chordate clades, recent work reveals conserved aspects of their
development, presumably inherited from the chordate common ancestor.
In amphioxus and vertebrates, the pharyngeal endoderm is specified by
attenuated retinoic acid signalling, and marked by conserved expression
of several transcription factors including pax1/9, six1/2, six4/5, six3/6,
eya, foxC and foxL1 (refs 32, 43, 44). In addition, recent work has shown
that the chordate ancestor probably had a collagen-based pharyngeal
skeleton incorporating cellular45 and acellular cartilage46,47 derived from
pharyngeal mesoderm. Whereas the pharyngeal walls develop pharyngeal
pouches and gill slits, the floor develops endostyle specializations related
to trapping food particles during filter feeding, as well as to hormonal
and protective functions.
The deuterostome roots of chordate characteristics
Work on hemichordates and echinoderms has informed our understand-
ing of ancestral deuterostome features, with different taxa contributing
complementary insights. Integrating insights from echinoderms is
challenging owing to the divergent radial body plan of adults, although
studies of echinoderm larval development have made essential contribu-
tions to our understanding of early deuterostome embryogenesis13,14,48,49.
Pterobranch hemichordates are relatively understudied50. In this Review,
therefore, we focus primarily on insights derived from the study of entero-
pneust hemichordates (acorn worms) as they relate to our understanding
of early deuterostome evolution.
As first described by Kowalevsky2, the anterior gut of hemichordates
is perforated in the dorsolateral region by a series of ciliated gill slits, now
known to be supported by gill bars composed of an acellular collagen
secreted by the endoderm (Fig. 2a, Box 1)46. Although there is no equiva-
lent structure in extant echinoderms, fossils reveal compelling evidence
that gill slits were present in stem echinoderms and subsequently lost51.
On the basis of morphological and functional criteria, enteropneust gill
slits closely resemble those of cephalochordates and are plausibly homolo-
gous1,46,52. In two species of enteropneust, studies of patterning genes with
conserved roles in chordate gill-pouch development, namely pax1/9, foxC,
foxL1, eya, six1 and foxI, also strongly support homology53–55.
The stomochord in hemichordates has drawn much comparative
interest as a notochord-like ancestral trait1,29,56,57. It is a diverticulum of
Ambulacrarians
Hemichordates Echinoderms Xenoturbelids
and acoelomorphs
and echinoderms, and chordates (blue), consisting of the cephalochordate,
tunicate and vertebrate lineages. Recent analyses have proposed either a
grouping of xenoturbellid and acoelomorph flatworms as sister group to
ambulacrarians27, or at the base of the bilaterians28(dashed lines).

a Proboscis Collar Trunk
Coeloms
A P
Pharynx
Mouth
Stomochord Collar cord Dorsal cord
Ventral cord
Figure 2 | Key anatomical features of the enteropneust body plan.
a, Longitudinal and transverse sections through an adult enteropneust
hemichordate, highlighting morphological characters that have featured
prominently in classic hypotheses of deuterostome evolution and chordate
the anterior gut that extends into the posterior proboscis supporting the
heart–kidney complex on its dorsal surface (Fig. 2a, Box 1). Stomochord
cells are vacuolated and surrounded by a sheath, similar in tissue organi-
zation to a notochord1,56,58. However, homology of these two structures
is weakly supported by both morphological and molecular criteria59–61.
In chordates the developing notochord is a key source of the secreted
BMP antagonists Chordin, Noggin and Follistatin, and the ventralizing
ligand Shh62. Of these genetic markers, only hh (the homologue of Shh)
is expressed in the stomochord, but it is also observed in surrounding
anterior endoderm57,63. Possible alternatives to notochord homology are
suggested by the stomochord expression of genes such as otx, dmbx, hex
and foxE that are expressed in prechordal endomesoderm of chordates,
but not in the notochord. These markers suggest that the stomochord is
an anterior endodermal structure with stronger affinities to the endostyle
than the notochord61.
The hemichordate nervous system is characterized by two contrasting
organizational features (Fig. 2b): a broad basiepithelial plexus, particularly
prominent in proboscis ectoderm, and a pair of nerve cords. The ventral
cord extends the length of the trunk and the dorsal cord runs from the
base of the proboscis down the length of the animal and joins to the ven-
tral cord by lateral nerve rings. Both cords are superficial condensations
of the nerve plexus except in a short length that spans the collar, where
the cord is internalized into a tube with a prominent lumen in some spe-
cies, and is formed by a developmental process that resembles chordate
neurulation59,64–67. Various authors have proposed both cords as possible
homologues of the chordate dorsal cord57,59,68,69, however, the internalized
collar cord has attracted the most attention6,57,59,69. Early reports suggested
that the dorsal cord was simply a through conduction tract of axons70,71.
Molecular studies, however, have shown condensations of cell bodies
associated with this cord6,69, and a further study in Balanoglossus simod-
ensis revealed bmp2/4, pax3/7 and msx expression in the collar cord57,
similar to that of the most lateral parts of the vertebrate neural plate and
in other bilaterians during CNS development. Although these similarities
are supportive of homology of the collar cord and chordate nerve cord,
other neural molecular markers complicate this interpretation. In Sac-
coglossus kowalevskii, markers of medial rather than lateral neural plate
are not expressed in the dorsal cord as predicted, but rather along the ven-
tral midline associated with the ventral cord. In addition, several neural
markers are not only expressed in the collar cord, but also throughout the
length of the superficial cord in the trunk, suggesting a patterning role
© 2015 Macmillan Publishers Limited. All rights reserved
REVIEW INSIGHT
b
Collar cord Dorsal
D cord
Dorsal
condensation
V
Ventral
cord
Gill slits Basiepithelial
nerve plexus
origins. A, anterior; P, posterior; D, dorsal; V, ventral. b, The nervous system of
an adult enteropneust showing both the broad basiepithelial plexus throughout
the ectoderm and nerve chords along the dorsal and ventral midlines. The blue
spots represent cell bodies and the lines represent neural processes.
throughout the dorsal midline72. When considering the general organiza-
tion of the nervous system in enteropneusts, no simple homology state-
ments can yet be made in relation to other nervous systems.
Although it seems likely that ancestral deuterostomes inherited some
elements of nervous system centralization from the bilaterian common
ancestor, a comprehensive characterization of key molecular markers is
needed to test competing hypotheses of nervous system evolution further.
It remains unclear whether the main features of the unusual enteropneust
nervous system can be ascribed to the filter-feeding deuterostome ances-
tor, thereafter modified in the chordate line, or whether they are second-
ary derivatives of the hemichordate lineage.
Axial patterning of deuterostome body plans
The discovery of conserved, pan-bilaterian mechanisms for the devel-
opment of the animal–vegetal, AP and DV body axes has transformed
our thinking about animal evolution4,5,73. This deep conservation initially
surprised biologists because of the great morphological diversity of bila-
terians, but made more sense when it was realized that the early axiation
processes of the embryo are separate from the later processes of morpho-
genesis, organogenesis and cell differentiation. Conserved suites of genes
are responsible for establishing basic regional differences of cells along all
three axes of bilaterian embryos, reflecting an extensive genetic regula-
tory network spread across the developing embryo. The resulting map of
conserved expression domains represents an ‘invisible anatomy’74 that
reveals clear relationships between disparate body plans, and provides a
window into the organization of expression domains in the deuterostome
ancestor. In this Review, we focus on the mechanisms by which these axes
are formed in deuterostomes, and the patterns of transcription-factor and
signalling-gene-expression domains produced along these axes.
The animal–vegetal axis and formation of endomesoderm
One of the first developmental decisions in embryogenesis is the estab-
lishment of the animal–vegetal axis. This axis sets up the formation of
the three germ layers: endoderm, mesoderm and ectoderm. Ectoderm
derives from the animal pole, and endomesoderm from the vegetal pole,
which later divides into endoderm and mesoderm. In all three major
deuterostome phyla, the formation of endomesoderm is triggered by
β-catenin protein, the intracellular effector of the canonical Wnt signalling
pathway. β-Catenin is stabilized preferentially in the vegetal pole of early
embryos and activates genes of the endomesodermal cellular program75,76.
2 3 A P R I L 2 0 1 5 | VO L 5 2 0 | NAT U R E | 4 5 9
 INSIGHT REVIEW
a Hemichordates Echinoderms Chordates
IsO
ZLI
D
ANR A P
?
V Anteroposterior axis
gbx
b
hox 1
hox 2, 3, 4
hox 5, 6, 7, 8
hox 9/10
hox11/13a, b, c
six3, bf-1, dlx, emx,
nkx2-1, rx, vax barH, tll en
lim1/5, otx, pax6
Fgf8 Sfrp1 Hh Wnt1
c nists are produced anteriorly from the mesoderm of Spemann’s organizer,
Dispersed Condensed
Figure 3 | A conserved molecular network for the deuterostome
anteroposterior axis.  a, Schematic representation of the distribution of
ectodermal expression domains of anteroposterior (AP) transcription factors
(blue gradient) and ectodermal signalling centres (green, yellow and red) in
relation to the body plans of deuterostome phyla. Chordate neuroectodermal
signalling centres depicted are the anterior neural ridge (ANR), zona limitans
intrathalamica (ZLI) and isthmic organizer (IsO). Broad conservation of
expression domains between hemichordates and chordates allows for the
reconstruction of an ancestral patterning network, which is shown without any
explicit inference of ancestral morphologies (b). Insufficient data exist from
echinoderms to infer to what extent they share this conserved AP patterning
network during adult patterning, although much of the anterior network is
conserved in larvae13,49. b, Domain map for the conserved transcription factors
and signalling ligands in relation to the AP axis63,85,86. c, Current data allow for
the reconstruction of a conserved molecular coordinate system for the AP axis
of the last common deuterostome ancestor, but not for the reconstruction of
discrete morphologies of that ancestor, because this AP patterning network is
deployed in a variety of morphological contexts, as evidenced by comparative
data from hemichordates (dispersed; AP expression domains encircling the
body) and chordates (condensed; AP domains largely restricted to regions near
the dorsal midline). A, anterior; P, posterior; D, dorsal; V, ventral.
In hemichordates and echinoderms, knockdown of the gene that encodes
β-catenin protein results in the ‘animalization’ of the embryo — excess
ectoderm and no endomesoderm. Conversely, stabilization of the protein
throughout the embryo results in ‘vegetalization’ of the embryo — excess
endomesoderm and no ectoderm. This mechanism has also been demon-
strated in protostomes from work on nemertine embryos77, and β-catenin
protein is also involved in endoderm formation in cnidarians78, suggesting
a deep eumetazoan ancestry for this process73.
Later specification of mesoderm from the endomesoderm occurs by
either of two generic mechanisms: autonomous specification by a cell’s
inheritance of a sequestered cytoplasmic determinant, or induction by a
signal from neighbouring tissue. In all deuterostomes except ascidians,
mesoderm formation occurs by induction. In vertebrates, two main sig-
nalling pathways are involved in mesoderm specification: Nodal and
FGF79. In amphioxus, FGF signalling specifies anterior mesoderm that
forms by enterocoely42. Similarly, in the hemichordate S. kowalevskii, FGF
signalling induces mesoderm and enterocoely, which raises the possibil-
ity of an ancestral role of FGF in deuterostome mesoderm formation80.
As a classic deuterostome character, a mechanistic link of enterocoely to
FGF signalling would support homology of this trait, at least within the
deuterostomes. In echinoderms, however, the role of FGF has yet to be
fully characterized, and there is some variation in inductive cues involved
in mesoderm specification: Notch–Delta signalling is important in early
mesoderm specification of echinoids, but not asteroids81. The differences
4 6 0 | NAT U R E | VO L 5 2 0 | 2 3 A P R I L 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved
between deuterostomes in specifying endomesoderm and mesoderm pre-
clude the definitive inference of the pathway of the deuterostome ancestor,
except that β-catenin protein is required at the start, and various inductive
signals are required later.
Although deuterostome taxa show an impressive array of morphologies,
organs and cell types along the AP axis, many of the early developmental
steps of axis formation are highly conserved and probably date back to the
bilaterian ancestor. Wnt signalling through β-catenin has emerged as the
earliest conserved determinant of AP pattern in deuterostomes. (Note that
this time and place of usage of β-catenin is separate from its role in endo-
mesoderm formation discussed earlier.) In vertebrates, Wnt proteins act
as posteriorizing signals in all three germ layers, but are most analysed in
CNS patterning82,83. Whereas Wnts are produced posteriorly, Wnt antago-
and their interaction sets up a graded Wnt distribution prefiguring the
eventual anatomical AP axis84. In both sea urchin larvae and the direct-
developing S. kowalevskii, Wnt signalling is also important for establishing
AP patterning48,63,75, suggesting that generating a Wnt signalling gradient
(high posteriorly, low anteriorly) is a key step in AP-axis formation in all
three phyla, for both adult and larval body plans. Different intensities of
Wnt signalling along the graded distribution then activate distinct genes
encoding different transcription factors and signalling ligands, produc-
ing a long-lasting AP map of gene expression domains that is collinear
with the Wnt distribution. The ectodermal map is strikingly similar in
the identity and relative expression of the constituent regulatory genes
across bilaterians85,86.
This conserved AP map provides a novel basis for comparing body
plans (Fig. 3a, b)74. In the most anterior regions, coexpression of genes
such as sfrp1/5, fgf8/17/18, foxG, retinal homeobox, dlx and nk2-1 define
ectodermal territories that later form proboscis ectoderm in hemi-
chordates and forebrain in vertebrates. Further posteriorly, expression
domains of emx, barH, dmbx and pax6 define the collar ectoderm of
hemichordates and midbrain of vertebrates; still more posteriorly,
domains of gbx, engrailed, pax2/5/8 and the collinearly expressed Hox
genes, regulate pharynx and trunk patterning of hemichordates and the
hindbrain and spinal cord in vertebrates (Fig. 3a)63,85,86. Enteropneust Hox
genes are organized as an intact cluster87, and in both cases the posterior
group Hox genes are expressed in post-anal parts of the body axis, perhaps
indicating domain-level homology of these deuterostomian posterior
appendages. AP map similarities even extend to three signalling centres,
producing the same signals and occupying equivalent map positions,
that are important for vertebrate brain patterning and for hemichordate
ectodermal development at the anterior tip, proboscis–collar boundary
and collar–trunk boundary (Fig. 3a,b) (for an alternative perspective see
ref. 88). In hemichordates, the conserved AP map of ectodermal expres-
sion domains covers both neural and epidermal tissue, and domains
encircle the body. In chordates most comparative studies have focused
on the role of this network in patterning the dorsal CNS, but more recent
studies demonstrate that expression of many of the genes extend ventrally
into sensory neurons and epidermis, suggesting a more general role in
ectodermal patterning88.
The AP map of expression domains provides a positional criterion for
evaluating morphological homologies between disparate body plans.
Thus, the homology of chordate and hemichordate gill slits is supported
by the observation that in both groups the first slit perforates the same
region of the AP expression map, near the midbrain–hindbrain boundary
in chordates and the collar–trunk boundary in enteropneusts. The map
also provokes comparisons: if the hemichordate collar cord is homologous
to the chordate dorsal nerve cord, it should express AP genes similar to
those of the chordate midbrain. We can confidently reconstruct this AP
patterning network in the ectoderm of the deuterostome ancestor, and as
previously mentioned, much of the map probably dates back to the bila-
terian ancestor. Indeed, more comparisons with protostomes are needed
to illuminate which few domains are deuterostome-unique, for example,

a Hemichordate
Ectoderm Dorsal
Mesoderm
midline
Bmp2/4
Ventral Bmp5-8
Endoderm
midline ADMP2
Neuralin
Chordin V D
Gremlin
ADMP
CV-2
Netrin
Tsg
Xolloid
Bambi
Saccoglossus Pinhead
Deuterostome
c ancestor
Dorsal
BMP
Inversion
Chordin
Ventral
Figure 4 | Comparison of the dorsoventral patterning mechanisms of
hemichordates and chordates.  a, BMP–Chordin signalling components
expressed in the dorsal and ventral midline ectoderm (blue) in the late
gastrula stage of Saccoglossus kowalevskii. b, BMP–Chordin signalling
components expressed either on the ventral side or dorsally in Spemann’s
in pharyngeal ectoderm and endoderm.
The AP axial homology of chordates and hemichordates with echi-
noderms is far less clear89–93. During the development of the larvae of
asteroids, echinoids and crinoids (Box 2), anterior regulatory genes are
expressed throughout the anterior ectoderm49,94, whereas posterior mark-
ers such as Hox genes are entirely absent during early patterning. In both
echinoids and crinoids, it is not until adult patterning begins in the late
larva that Hox genes begin expression in a collinear pattern, not in ecto-
derm, but in posterior coelomic mesoderm90,95. Some anterior markers
are expressed in the oral ectoderm and tube feet of pentaradial adults,
but current data are too fragmentary to make valid comparisons of adult
echinoderms with other deuterostome adults. Comprehensive characteri-
zation of the patterning of echinoderm adults is badly needed to unravel
the evolution of this unusual body plan.
Although the AP map is conserved across deuterostomes (and in most
aspects, across bilaterians) the differentiated morphologies built on it are
probably not (Fig. 3c). The morphological outcomes of development dif-
fer in each phylum because the transcription factors and signals of the
conserved map activate and repress different target genes63. These target
genes, in turn, direct the final steps of organogenesis, morphogenesis and
cell-type formation.
The dorsoventral dimension
The DV axis evolved on the Pre-Cambrian stem leading to the bilaterian
ancestor, and is intimately tied with the origin of bilateral symmetry. Its
formation in early embryogenesis is analogous to AP axis formation. One
midline of the embryo produces Bmp, and the opposite midline pro-
duces the Bmp antagonist Chordin96. Through complex interactions, this
antagonism generates a graded distribution of Bmp across the embryo, a
graded occupancy of Bmp receptors, and a corresponding graded distri-
bution of activated Smad1/5 transcription factor in embryonic cells. This
gradient of activated Smad1/5 stimulates and represses different genes
encoding transcription factors and other signalling ligands, generating a
long-lasting DV map of expression domains of these genes5,97 (Fig. 4a, b).
The patterns of transcription-factor and signalling-ligand expression
established along the DV direction generate the corresponding anatomical
© 2015 Macmillan Publishers Limited. All rights reserved
REVIEW INSIGHT
Spemann’s
b Chordate
organizer
CNS Chordin
Ventral
centre Noggin
Follistatin
CV-2
ADMP
Bmp4
Frzb-1
Tsg V D
Crescent
Xlr
sFRP2
Xolloid
Dkk-1
Bambi
Cerberus
Pinhead
Antivin/Lefty
Sizzled
Xnrs
Xenopus
Shh
IGFBP5
Chordate
ancestor
Dorsal
Chordin
BMP
Ventral
organizer in the early gastrula of Xenopus. CNS, central nervous system. c,
The inversion of dorsoventral (DV) signalling centres and the relocation
of the Chordin source from the ectoderm (yellow) to mesoderm (red)
were innovations in DV patterning at the base of the chordates (ancestral
location shown by grey shading).
axis by driving the expression of genes for the development of different tis-
sues and cell types in different regions. Some of the definitive tissues and
cell types are remarkably conserved among bilaterians, as demonstrated
by the similarities between the DV development of protostomes such as
the fruit fly and the annelid Platynereis dumerilii and vertebrates such
as Xenopus, mice and zebrafish97,98. Domains from the Chordin side of
the Bmp distribution activate axial (striated) muscle development in the
mesoderm and nerve-cell development in the ectoderm, especially motor
neurons and interneurons that assemble into the CNS, whereas domains
from the Bmp side activate heart tube and coelom development from the
mesoderm and epidermis and sensory-nerve-cell development from the
ectoderm97. The Bmp distribution patterns all three germ layers.
Although deuterostomes as a group inherited the basic mechanism of
DV axis formation from the bilateral ancestor, there are important differ-
ences among them that can inform hypotheses of chordate origins. It is
immediately apparent that the Chordin and Bmp sides of the molecular
DV axis have different anatomical names in deuterostomes and protos-
tomes. In deuterostomes, the Bmp side is called ‘dorsal’ and the Chordin
side is ‘ventral’, but in Drosophila and other protostomes the molecular
and anatomical links are reversed. By zoological convention, sides are
named according to the animal’s orientation to the substratum and the
location of the mouth. The difference was resolved by the proposal that
the chordate ancestor underwent a dorsoventral inversion of the body
relative to the substratum. This transition simultaneously inverted the
Bmp–Chordin axis, the domain map, and axis of anatomical differentia-
tions5,96. As a final refinement the mouth was relocated to the Bmp side,
whereas most protostomes (for example, Drosophila), and invertebrate
deuterostomes, form the mouth on the Chordin side. Although seeming
modest as a novelty, body inversion must be considered when discussing
innovations of the chordate line.
S. kowalevskii provides an excellent example of bilaterian DV axiation,
probably conserved from the deuterostome ancestor72. (Indirect develop-
ing hemichordates and echinoderms also exhibit Bmp–Chordin-based
DV patterning, modified for larval body plans, although we cannot cover
these here9,99.) At gastrula stages, bmp2/4 is strongly expressed on the dor-
sal ectodermal midline of S. kowalevskii, accompanied by genes for a large
2 3 A P R I L 2 0 1 5 | VO L 5 2 0 | NAT U R E | 4 6 1
 INSIGHT REVIEW
BOX 2
Deuterostome larval diversity
a, Hemichordates and echinoderms include lineages that are
characterized by both direct-developers, forming the adult body plan
from embryogenesis in a matter of days; and by indirect-developers,
first forming planktonic feeding larvae that may swim and feed for
months before metamorphosing to produce a distinct adult body
plan. The harrimaniid enteropneusts and pterobranchs (reproduced
from ref. 49) are examples of direct-developers. b, A model of a
two-day-old embryo of Saccoglossus kowalevskii contrasts with the
month-old late spengelid tornaria larva. c, The organization of the
tornaria larval body plan is very similar to the organization of the
echinoderm larva represented here by a holothuroid auricularian
larva. However, echinoderms have a spectacular variety of larval forms
from the ophiopluteus and echinopluteus with similar elaborate
skeletons to the asteroid bipinnarian and holthuroid auricularian
larvae with similar convoluted ciliary bands. Many researchers have
focused on the morphological and developmental similarities between
the diverse ambulacrarian larval types, suggesting the existence of
an ancestral ‘dipleurula’ (small two-sided) larval form from which
ambulacrarian larval diversity arose114. The dipleurula ectoderm is
a Hemichordates
Pterobranchs Spengelids Harrimaniids Crinoids
b c
Eye spots
Gill slits Apical ganglion
Mouth Protocoel
Heart–kidney
complex
Mesocoels
Ciliary
Metacoels
band
Stomach
Telotroch
Anus
Indirect development Direct development
(tornaria larva) (S. kowalevskii)
set of signal modulating proteins and other Bmp-related proteins. Con-
versely, chordin and admp are strongly expressed on the opposite, ventral
midline (Fig. 4a). Following the Bmp distribution gradient, transcription-
factor genes are activated in a DV map that generally parallels the expres-
sion of orthologous genes in Drosophila and vertebrates. The DV domain
map and subsequent differentiated structures of the overt anatomical axis
depend entirely on the Bmp distribution, as shown by the development
of dorsalized embryos in the presence of excess uniform Bmp2/4 protein,
and of ventralized embryos when Bmp2/4 is eliminated72. Tissues, organs
and cell types of the three germ layers are patterned by the Bmp–Chordin
distribution, including the gill slits, the mouth and the two nerve cords
(Fig. 2a). In embryos dorsalized by excessive Bmp, nerve cells still form
in abundance. Although this might seem contrary to chordate neural
4 6 2 | NAT U R E | VO L 5 2 0 | 2 3 A P R I L 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved
characterized by a convoluted ciliary band used for swimming and
feeding. In hemichordates, a robust additional posterior band of
compound cilia, the telotroch, is purely locomotory (a, b). The nervous
system is divided into two domains: an apical ganglion underlying
the sensory ciliated apical organ, and neurons underlying the length
of the ciliary bands (b, c). The dipleurula mesoderm is formed by
enterocoely and organized into three compartments: anterior, middle
and posterior. In echinoderms, the adult body plan is initiated by
the left middle coelom, which expands and forms five lobes midway
through larval development (b, asterisk in c). An influential theory of
Garstang108 further elaborated by a variety of authors (see Review
by Holland et al. on page 450), proposed that the deuterostome
ancestor also had a dipleurula larva, and that chordates evolved by
paedomorphosis from such forms. A central tenet of this theory is
that the dorsal central nervous system of chordates evolved through
the dorsal migration and fusion of the lateral ciliary bands of the
dipleurula larvae, and their underlying neurons114. More recently, this
hypothesis has fallen out of favour on the basis of both phylogenetic
and body-patterning data21,24,109.
Echinoderms
Asteroids Ophiuroids Holothuroids Echinoids
Ciliary band
Apical
ganglion
Axocoels
Mouth
Hydrocoels
*
Stomach
Anus
Somatocoels
Ciliary band
neurons
Early larva Late larva
patterning in which Bmp initially represses neural development in the
epidermis, it is not; the hemichordate dorsal nerve cord normally forms
at the midline of high Bmp concentration, and the lateral parts of the
chordate neural plate are themselves patterned by high Bmp concentra-
tions. Overall, the hemichordate findings affirm general insights about
bilaterian DV axis formation. In its body orientation, S. kowalevskii resem-
bles protostomes: Bmp foretells the ventral side and Chordin the dorsal,
leaving chordates as the single ‘inverted’ phylum (Fig. 4c). One of the
key questions about chordate origins remains the evolution of the dorsal
hollow nerve cord from the nervous system of a less centralized ancestor
with little or no capacity for neurulation. In general, hypotheses imply
that in the early embryo, the formation of neural ectoderm (prospective
for motor neurons and interneurons) was increasingly repressed towards

one midline, and the neurulation process was induced along the edges of
the narrowed neurectoderm territory. Thereafter Bmp exerted its neural
patterning effects from the neural plate borders. This, of course, remains
an area for future investigation6,68,100.
Chordates differ from hemichordates in that Chordin and other Bmp
antagonists are produced mostly in midline mesoderm, and specifically
in mesoderm of Spemann’s organizer, a region formed in the late blastula
embryo at a location of high Nodal signalling and low Bmp signalling101.
Organizer cells are precursors of the notochord and head mesoderm.
Notochord precursors undergo extreme convergent extension by cell
intercalation, forcefully repacking a cube of cells into a rod one-cell wide
and lengthening the embryonic midline. Simultaneously they secrete
their dorsoventral patterning molecules, neuralizing nearly half the
embryo’s overlying ectoderm and initiating neurulation morphogenesis.
In chordates, neurogenic ectoderm produces little or no Bmp antago-
nist, whereas in hemichordates it is the main source. Recently, a possible
notochord homologue, the axochord, was described for the polychaete
annelid P. dumerilii; it is a midline mesodermal structure of muscle cells
contained in a strong sheath into which lateral muscles attach, but there
is no evidence that it is a notochord-like signalling source102. Rather, the
midline signalling source of this protostome is presumably nearby neu-
ral ectoderm that determines, among other things, where the axochord
itself develops.
Taken together, these data suggest that hemichordates are like proto-
stomes in their dorsoventral development, whereas chordates have con-
siderably modified the ancestral patterning mode (Fig. 4c), by adding
organizer mesoderm as the Chordin source, and acquiring a large-scale
neuralization response in the ectoderm. These innovations in DV axiation
must be considered in any discussion of chordate origins.
Gill slits are a deuterostome innovation
The single unambiguous anatomical homology that is a clear deuteros-
tome synapomorphy is the pharyngeal gill-slit complex6,46,53,55,60,103. These
perforations of pharyngeal endoderm and ectoderm, ringed by beating
cilia, imply that the ancestor fed by ingesting food particles carried by
water flow entering the mouth and exiting the slits. The complex, which
can include more than a hundred (bilaterally symmetrical) gill-slit pairs,
is a major developmental and morphological modification beyond the
bilaterian ancestor’s pharynx, although presumably elaborated from it.
Although hemichordates do not have a well-defined pharyngeal endostyle
like chordates, the pharynx as a whole, and even the proboscis, probably
makes endostyle-like mucociliary contributions to food trapping and
conveyance to the gut52,60. Some of these functions may be deuterostome
synapomorphies. To coordinate the functions of gill-slit-mediated water
propulsion, food intake, trapping and conveyance, the pharyngeal nerv-
ous system is likely to have become modified from that of the bilateral
ancestor. Given that pharyngeal innovations may represent the signature
morphological, developmental and genomic innovations of deuteros-
tomes, their development and physiology should be characterized more
comprehensively.
Among extant animals, the filter-feeding lifestyle correlates with simpli-
fied body plans — radialized dorsoventral dimensions, more dispersed
nervous systems, less cephalization of sensory systems, and less motility by
trunk and tail axial muscles — when compared with extant food-seeking or
predatory arthropods, annelids and jawed vertebrates. Such simplifications
are presumably anatomical or physiological adaptations that benefit gill-
slit-mediated filter feeding, and it seems plausible that evolution along the
deuterostome stem involved considerable morphological modifications
relative to earliest bilaterian body plans. If true, it is nonetheless apparent
from AP and DV domain maps that the deuterostome ancestor suffered no
concomitant loss of body-plan complexity at the molecular genetic devel-
opmental level. Rather, it shows that bilaterian domain maps are remark-
ably stable and can support wide-ranging morphologies, organogenesis
and cytodifferentiations. An example of such modification is the muscular
proboscis of hemichordates. The proboscis is used to dig and to trap food,
while containing most of the conserved basic patterning elements of the
© 2015 Macmillan Publishers Limited. All rights reserved
REVIEW INSIGHT
vertebrate forebrain, here spread over a basiepithelial nerve plexus63,85.
There would be no intrinsic reason for the deuterostome ancestor to pre-
serve the morphology and differentiations of the bilaterian ancestor if it
no longer lived that ancestor’s lifestyle. Finally, there is some palaeonto-
logical support for a filter-feeding deuterostome ancestor. On the basis
of molecular clock estimates, deuterostome phyla would have diverged
in the Ediacaran period, well before the Cambrian explosion. The lack of
an obvious fossil record, except for small Precambrian trace fossils and
the enigmatic Ediacaran fossils, and evidence of abundant filterable food
sources in the form of microbial mats and plankton, suggest that bilaterians
of that time were probably small and simple filter feeders104–106.
This interpretation of the deuterostome ancestor has important con-
sequences for the origin of chordates. Relative to that ancestor, the chor-
date stem lineage achieved major developmental and morphological
innovations, including the evolution of a true notochord from the arch-
enteron roof, centralizing many morphogenetic activities of the ances-
tral archenteron and taking over the signalling activities of the ancestral
ectoderm for both AP and DV axial patterning (by producing Bmp and
Wnt antagonists), to become the centrepiece of Spemann’s organizer.
Concomitantly, the innovations of neural induction (neuralization) and
full-length neurulation of the ectoderm generated a hollow nerve cord
along the entire body length (a length now defined by the elongating
notochord), rather than just the short and late collar cord neurulation of
hemichordates (although this limited neurulation shows that the ancestor
possessed the basic morphogenetic process and components). At some
point later, dorsoventral inversion of the chordate body took place, with
mouth relocation out of the neural ectoderm107. All of this occurred on
the chordate stem, perhaps after the elimination of an ambulacrarian-type
larva, to open up uninterrupted embryonic development of the adult body
plan (see ref. 100 for further elaboration of this hypothesis).
Future directions
Insights into deuterostome evolution are emerging from research in
developmental biology, phylogenomics, genomics and zoology. A par-
ticular focus has been the pharyngeal gill-slit complex, which is sup-
ported as an ancestral deuterostome feature by strong morphological
and developmental data. The implication that the deuterostome ances-
tor was a filter feeder naturally draws attention to other integrated phar-
yngeal specializations, including endostyle-like food-trapping organs.
Further study of these organs, especially in amphioxus and hemichor-
dates, has the immediate potential to reveal clues about deuterostome
and chordate origins.
One of the most important differences between hemichordates
and chordates, revealed by comparative developmental studies, is the
source of Bmp antagonists involved in establishing DV axial polarity
in early development. These antagonists are expressed in the ectoderm
of hemichordates and the mesoderm of the chordate organizer. Spe-
mann’s organizer is a key chordate developmental innovation defined
by various secreted factors modulating Bmp, Nodal and Wnt signalling.
A more comprehensive description of the roles of these signals and their
antagonists in the patterning of hemichordate mesoderm and ectoderm
will be required to devise and test hypotheses about the evolution of the
chordate organizer.
Most developmental insights from hemichordates have so far come
from studies of direct developing hemichordates, but a distinct larval
life-history stage is probably an ancestral trait of Ambulacraria and per-
haps of deuterostomes (Box 2). More comprehensive developmental
studies in indirect-developing echinoderms and hemichordates, with
distinct larval body plans, are needed to determine the importance of
complex life cycles and the role of larvae in the early diversification of
deuterostome body plans. Garstang’s influential auricularian hypothesis
derived the chordate body plan from an ancestral larval body plan108,
but this hypothesis has recently lost support due to revisions in chor-
date phylogeny and close similarities between adult rather than larval
body patterning109. Comparative data sets on larval patterning will
be key for reconstructing ancestral developmental strategies of early
2 3 A P R I L 2 0 1 5 | VO L 5 2 0 | NAT U R E | 4 6 3
 INSIGHT REVIEW
deuterostomes and testing hypotheses of larval homology.
Finally, advances in genomics have begun to shed light on the gene
content and chromosomal organization of invertebrate deuterostomes,
including the purple sea urchin (Strongylocentrotus purpuratus)110, the
acorn worms S. kowalevskii and Ptychodera flava87,111, and the crown-of-
thorns sea star (Acanthaster planci)112. Given the apparent conservation
of not only the pan-deuterostome axial maps but also many of the down-
stream factors that control organogenesis, it will be exciting to explore the
gene-regulatory elements that underlie this deep conservation through
a combination of comparative genomics and experimental developmen-
tal biology, revealing features of the ancestral deuterostome down to the
nucleotide level. ■
Received 24 October 2014; accepted 3 February 2015.
1. Bateson, W. The ancestry of the chordata. Q. J. Microsc. Sci. 26, 535–571
(1886).
2. Kowalevsky, A. Anatomie des Balanoglossus [in French]. Mem. Acad. Imp. Sci. St
Petersb. 7, 16 (1866).
3. Metchnikoff, V. E. Über die systematische Stellung von Balanoglossus [in
German]. Zool. Anz. 4, 139–157 (1881).
4. Carroll, S. B., Grenier, J. K. & Weatherbee, S. D. From DNA to Diversity: Molecular
Genetics and the Evolution of Animal Design 2nd edn (Blackwell, 2005).
5. De Robertis, E. M. Evo-devo: variations on ancestral themes. Cell 132, 185–195
(2008).
6. Brown, F. D., Prendergast, A. & Swalla, B. J. Man is but a worm: chordate origins.
Genesis 46, 605–613 (2008).
7. Gerhart, J., Lowe, C. & Kirschner, M. Hemichordates and the origin of chordates.
Curr. Opin. Genet. Dev. 15, 461–467 (2005).
8. Holland, L. Z. Chordate roots of the vertebrate nervous system: expanding the
molecular toolkit. Nature Rev. Neurosci. 10, 736–746 (2009).
9. Lapraz, F., Besnardeau, L. & Lepage, T. Patterning of the dorsal-ventral axis in
echinoderms: insights into the evolution of the BMP-chordin signaling network.
PLoS Biol. 7, e1000248 (2009).
10. Satoh, N. Developmental Genomics of Ascidians (Wiley-Blackwell, 2014).
11. Yu, J. K. et al. Axial patterning in cephalochordates and the evolution of the
organizer. Nature 445, 613–617 (2007).
This article reports molecular evidence for a functional organizer in
amphioxus, and supports the presence of this molecular patterning module at
the base of the chordates.
12. Röttinger, E. & Lowe, C. J. Evolutionary crossroads in developmental biology:
hemichordates. Development 139, 2463–2475 (2012).
13. Angerer, L. M., Yaguchi, S., Angerer, R. C. & Burke, R. D. The evolution of nervous
system patterning: insights from sea urchin development. Development 138,
3613–3623 (2011).
14. Peter, I. S. & Davidson, E. H. A gene regulatory network controlling the
embryonic specification of endoderm. Nature 474, 635–639 (2011).
15. Swalla, B. J. & Smith, A. B. Deciphering deuterostome phylogeny: molecular,
morphological and palaeontological perspectives. Phil. Trans. R. Soc. Lond. B
363, 1557–1568 (2008).
16. Haeckel, E. H. P. A. Anthropogenie: oder, Entwickelungsgeschichte des Menschen
[in German] (Wilhelm Engelmann, 1874).
17. Kowalevsky, A. Weitere Studien über die Entwicklung der einfachen Ascidien [in
German]. Arch. Mikr. Anat. 7, 101–130 (1866).
18. Darwin, C. The Descent of Man, and Selection in Relation to Sex (D. Appleton and
company, 1871).
19. Grobben, K. Die systematische einteilung des tierreiches [in German]. Verh. der
Zool.-Bot. Ges. Wien. 58, 491–511 (1908).
20. Martin-Duran, J. M., Janssen, R., Wennberg, S., Budd, G. E. & Hejnol, A.
Deuterostomic development in the protostome Priapulus caudatus. Curr. Biol.
22, 2161–2166 (2012).
21. Delsuc, F., Brinkmann, H., Chourrout, D. & Philippe, H. Tunicates and not
cephalochordates are the closest living relatives of vertebrates. Nature 439,
965–968 (2006).
The authors of this paper provide conclusive molecular evidence for the sister
relationship between tunicates and vertebrates and the basal position of
cephalochordates.
22. Dunn, C. W. et al. Broad phylogenomic sampling improves resolution of the
animal tree of life. Nature 452, 745–749 (2008).
23. Wada, H. & Satoh, N. Details of the evolutionary history from invertebrates to
vertebrates, as deduced from the sequences of 18S rDNA. Proc. Natl Acad. Sci.
USA 91, 1801–1804 (1994).
24. Blair, J. E. & Hedges, S. B. Molecular phylogeny and divergence times of
deuterostome animals. Mol. Biol. Evol. 22, 2275–2284 (2005).
25. Field, K. G. et al. Molecular phylogeny of the animal kingdom. Science 239,
748–753 (1988).
26. Halanych, K. M. The phylogenetic position of the pterobranch hemichordates
based on 18S rDNA sequence data. Mol. Phylogenet. Evol. 4, 72–76 (1995).
27. Philippe, H. et al. Acoelomorph flatworms are deuterostomes related to
Xenoturbella. Nature 470, 255–258 (2011).
28. Hejnol, A. et al. Assessing the root of bilaterian animals with scalable
phylogenomic methods. Proc. Biol. Sci. 276, 4261–4270 (2009).
29. Gee, H. Before the Backbone (Chapman & Hall, 1996).
30. Stach, T. Chordate phylogeny and evolution: a not so simple three-taxon
4 6 4 | NAT U R E | VO L 5 2 0 | 2 3 A P R I L 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved
problem. J. Zool. 276, 117–141 (2008).
31. Lu, T. M., Luo, Y. J. & Yu, J. K. BMP and Delta/Notch signaling control the
development of amphioxus epidermal sensory neurons: insights into the
evolution of the peripheral sensory system. Development 139, 2020–2030
(2012).
32. Escriva, H., Holland, N. D., Gronemeyer, H., Laudet, V. & Holland, L. Z. The
retinoic acid signaling pathway regulates anterior/posterior patterning in
the nerve cord and pharynx of amphioxus, a chordate lacking neural crest.
Development 129, 2905–2916 (2002).
33. Onai, T. et al. Retinoic acid and Wnt/β-catenin have complementary roles in
anterior/posterior patterning embryos of the basal chordate amphioxus. Dev.
Biol. 332, 223–233 (2009).
34. Meulemans, D. & Bronner-Fraser, M. Gene-regulatory interactions in neural
crest evolution and development. Dev. Cell 7, 291–299 (2004).
35. Shimeld, S. M. The evolution of the hedgehog gene family in chordates: insights
from amphioxus hedgehog. Dev. Genes Evol. 209, 40–47 (1999).
36. Holland, L. Z. et al. Evolution of bilaterian central nervous systems: a single
origin? EvoDevo 4, 27 (2013).
37. Irimia, M. et al. Conserved developmental expression of Fezf in chordates
and Drosophila and the origin of the Zona Limitans Intrathalamica (ZLI) brain
organizer. EvoDevo 1, 7 (2010).
38. Abedin, M. & King, N. The premetazoan ancestry of cadherins. Science 319,
946–948 (2008).
39. Conklin, E. G. The embryology of amphioxus. J. Morphol. 54, 69–151 (1932).
40. Gomez, C. et al. Control of segment number in vertebrate embryos. Nature 454,
335–339 (2008).
41. Beaster-Jones, L. et al. Expression of somite segmentation genes in amphioxus:
a clock without a wavefront? Dev. Genes Evol. 218, 599–611 (2008).
42. Bertrand, S. et al. Amphioxus FGF signaling predicts the acquisition of
vertebrate morphological traits. Proc. Natl Acad. Sci. USA 108, 9160–9165
(2011).
43. Kozmik, Z. et al. Pax-Six-Eya-Dach network during amphioxus development:
conservation in vitro but context specificity in vivo. Dev. Biol. 306, 143–159
(2007).
44. Mazet, F., Amemiya, C. T. & Shimeld, S. M. An ancient Fox gene cluster in
bilaterian animals. Curr. Biol. 16, R314–316 (2006).
45. Jandzik, D. et al. Evolution of the new vertebrate head by co-option of an ancient
chordate skeletal tissue. Nature 518, 534–537 (2015).
46. Rychel, A. L. & Swalla, B. J. Development and evolution of chordate cartilage.
J. Exp. Zool. B Mol. Dev. Evol. 308, 325–335 (2007).
47. Wright, G. M., Keeley, F. W. & Robson, P. The unusual cartilaginous tissues of
jawless craniates, cephalochordates and invertebrates. Cell Tissue Res. 304,
165–174 (2001).
48. Range, R. C., Angerer, R. C. & Angerer, L. M. Integration of canonical and
noncanonical Wnt signaling pathways patterns the neuroectoderm along the
anterior–posterior axis of sea urchin embryos. PLoS Biol. 11, e1001467 (2013).
49. Yankura, K. A., Martik, M. L., Jennings, C. K. & Hinman, V. F. Uncoupling of
complex regulatory patterning during evolution of larval development in
echinoderms. BMC Biol. 8, 143 (2010).
50. Sato, A., Bishop, J. D. & Holland, P. W. Developmental biology of pterobranch
hemichordates: history and perspectives. Genesis 46, 587–591 (2008).
51. Dominguez, P., Jacobson, A. G. & Jefferies, R. P. Paired gill slits in a fossil with a
calcite skeleton. Nature 417, 841–844 (2002).
52. Gonzalez, P. & Cameron, C. B. The gill slits and pre-oral ciliary organ of
Protoglossus (Hemichordata: Enteropneusta) are filter-feeding structures. Biol. J.
Linn. Soc. 98, 898–906 (2009).
53. Gillis, J. A., Fritzenwanker, J. H. & Lowe, C. J. A stem-deuterostome origin of the
vertebrate pharyngeal transcriptional network. Proc. Biol. Sci. 279, 237–246
(2012).
This report shows extensive patterning similarities between the development
of chordate and enteropneust gills, further supporting morphological
homology.
54. Ogasawara, M., Wada, H., Peters, H. & Satoh, N. Developmental expression of
Pax1/9 genes in urochordate and hemichordate gills: insight into function and
evolution of the pharyngeal epithelium. Development 126, 2539–2550 (1999).
55. Fritzenwanker, J. H., Gerhart, J., Freeman, R. M. Jr & Lowe, C. J. The Fox/
Forkhead transcription factor family of the hemichordate Saccoglossus
kowalevskii. EvoDevo 5, 17 (2014).
56. Balser, E. J. & Ruppert, E. E. Structure, ultrastructure, and function of the preoral
heart-kidney in Saccoglossus kowalevskii (Hemichordata, Enteropneusta)
including new data on the stomochord. Acta Zool. 71, 235–249 (1990).
57. Miyamoto, N. & Wada, H. Hemichordate neurulation and the origin of the neural
tube. Nature Commun. 4, 2713 (2013).
This manuscript demonstrates similar mediolateral patterning mechanisms
between the hemichordate collar cord and chordate dorsal cord.
58. Hyman, L. H. The Invertebrates 1st edn (McGraw-Hill, 1959).
59. Luttrell, S., Konikoff, C., Byrne, A., Bengtsson, B. & Swalla, B. J. Ptychoderid
hemichordate neurulation without a notochord. Integr. Comp. Biol. 52, 829–834
(2012).
60. Ruppert, E. E. Key characters uniting hemichordates and chordates:
homologies or homoplasies? Can. J. Zool. 83, 8–23 (2005).
61. Satoh, N. et al. On a possible evolutionary link of the stomochord of
hemichordates to pharyngeal organs of chordates. Genesis 52, 925–934
(2014).
62. De Robertis, E. M. & Kuroda, H. Dorso-ventral patterning and neural induction in
Xenopus embryos. Annu. Rev. Cell Dev. Biol. 20, 285–308 (2004).

63. Pani, A. M. et al. Ancient deuterostome origins of vertebrate brain signalling
centres. Nature 483, 289–294 (2012).
This paper presents evidence that ectodermal signalling centres thought
to have been uniquely associated with the evolution of vertebrate brains
are present in hemichordates as part of a conserved ancient deuterostome
patterning network.
64. Bullock, T. H. The anatomical organization of the nervous system of
enteropneusta. Q. J. Microsc. Sci. 86, 55–111 (1945).
65. Kaul, S. & Stach, T. Ontogeny of the collar cord: neurulation in the hemichordate
Saccoglossus kowalevskii. J. Morphol. 271, 1240–1259 (2010).
66. Knight-Jones, E. On the nervous system of Saccoglossus cambriensis
(Enteropneusta). Phil. Trans. R. Soc. Lond. B 236, 315–354 (1952).
67. Morgan, T. Development of Balanoglossus. J. Morphol. 9, 1–86 (1894).
68. Benito-Gutierrez, E. & Arendt, D. CNS evolution: new insight from the mud. Curr.
Biol. 19, R640–642 (2009).
69. Nomaksteinsky, M. et al. Centralization of the deuterostome nervous system
predates chordates. Curr. Biol. 19, 1264–1269 (2009).
This paper shows clear molecular evidence for the presence of cell bodies in
the dorsal nerve cord of enteropneusts and proposes the deep ancestry of a
CNS in the deuterostomes.
70. Bullock, T. H. The functional organisation of the nervous system of the
Enteropneusta. Biol. Bull. 79, 91–113 (1940).
71. Cameron, C. B. & Mackie, G. O. Conduction pathways in the nervous system of
Saccoglossus sp. (Enteropneusta). Can. J. Zool. 74, 15–19 (1996).
72. Lowe, C. J. et al. Dorsoventral patterning in hemichordates: insights into early
chordate evolution. PLoS Biol. 4, e291 (2006).
This manuscript demonstrates the role of BMP signalling in the formation of
the DV axis of the enteropneust adult body plan.
73. Martindale, M. Q. & Hejnol, A. A developmental perspective: changes in the
position of the blastopore during bilaterian evolution. Dev. Cell 17, 162–174
(2009).
74. Slack, J. M., Holland, P. W. & Graham, C. F. The zootype and the phylotypic stage.
Nature 361, 490–492 (1993).
75. Darras, S., Gerhart, J., Terasaki, M., Kirschner, M. & Lowe, C. J. β-Catenin
specifies the endomesoderm and defines the posterior organizer of the
hemichordate Saccoglossus kowalevskii. Development 138, 959–970 (2011).
76. Wikramanayake, A. H., Huang, L. & Klein, W. H. β-Catenin is essential for
patterning the maternally specified animal-vegetal axis in the sea urchin
embryo. Proc. Natl Acad. Sci. USA 95, 9343–9348 (1998).
77. Henry, J. Q., Perry, K. J., Wever, J., Seaver, E. & Martindale, M. Q. β-Catenin is
required for the establishment of vegetal embryonic fates in the nemertean,
Cerebratulus lacteus. Dev. Biol. 317, 368–379 (2008).
78. Wikramanayake, A. H. et al. An ancient role for nuclear β-catenin in the evolution
of axial polarity and germ layer segregation. Nature 426, 446–450 (2003).
79. Kimelman, D. Mesoderm induction: from caps to chips. Nature Rev. Genet. 7,
360–372 (2006).
80. Green, S. A., Norris, R. P., Terasaki, M. & Lowe, C. J. FGF signaling induces
mesoderm in the hemichordate Saccoglossus kowalevskii. Development 140,
1024–1033 (2013).
81. Hinman, V. F. & Davidson, E. H. Evolutionary plasticity of developmental gene
regulatory network architecture. Proc. Natl Acad. Sci. USA 104, 19404–19409
(2007).
82. Hikasa, H. & Sokol, S. Y. Wnt signaling in vertebrate axis specification. Cold
Spring Harb. Perspect. Biol. 5, a007955 (2013).
83. Petersen, C. P. & Reddien, P. W. Wnt signaling and the polarity of the primary
body axis. Cell 139, 1056–1068 (2009).
84. Kiecker, C. & Niehrs, C. A morphogen gradient of Wnt/β-catenin signalling
regulates anteroposterior neural patterning in Xenopus. Development 128,
4189–4201 (2001).
85. Lowe, C. J. et al. Anteroposterior patterning in hemichordates and the origins of
the chordate nervous system. Cell 113, 853–865 (2003).
This paper provides evidence for a conserved transcriptional gene regulatory
network between hemichordates and chordates despite large organizational
differences in their basic body plans.
86. Aronowicz, J. & Lowe, C. J. Hox gene expression in the hemichordate
Saccoglossus kowalevskii and the evolution of deuterostome nervous systems.
Integr. Comp. Biol. 46, 890–901 (2006).
87. Freeman, R. et al. Identical genomic organization of two hemichordate hox
clusters. Curr. Biol. 22, 2053–2058 (2012).
88. Holland, L. Z. et al. Evolution of bilaterian central nervous systems: a single
origin? EvoDevo 4, 27 (2013).
89. David, B. & Mooi, R. How Hox genes can shed light on the place of echinoderms
among the deuterostomes. EvoDevo 5, 22 (2014).
90. Hara, Y. et al. Expression patterns of Hox genes in larvae of the sea lily
Metacrinus rotundus. Dev. Genes Evol. 216, 797–809 (2006).
91. Morris, V. B. & Byrne, M. Oral-aboral identity displayed in the expression of
HpHox3 and HpHox11/13 in the adult rudiment of the sea urchin Holopneustes
purpurescens. Dev. Genes Evol. 224, 1–11 (2014).
92. Lacalli, T. Echinoderm conundrums: Hox genes, heterochrony, and an excess of
mouths. Evodevo 5, 46 (2014).
93. Peterson, K. J., Arenas-Mena, C. & Davidson, E. H. The A/P axis in echinoderm
ontogeny and evolution: evidence from fossils and molecules. Evol. Dev. 2,
93–101 (2000).
94. Omori, A., Akasaka, K., Kurokawa, D. & Amemiya, S. Gene expression analysis of
Six3, Pax6, and Otx in the early development of the stalked crinoid Metacrinus
rotundus. Gene Expr. Patterns 11, 48–56 (2011).
95. Arenas-Mena, C., Cameron, A. R. & Davidson, E. H. Spatial expression of Hox
© 2015 Macmillan Publishers Limited. All rights reserved
REVIEW INSIGHT
cluster genes in the ontogeny of a sea urchin. Development 127, 4631–4643
(2000).
96. De Robertis, E. M. & Sasai, Y. A common plan for dorsoventral patterning in
Bilateria. Nature 380, 37–40 (1996).
97. Mizutani, C. M. & Bier, E. EvoD/Vo: the origins of BMP signalling in the
neuroectoderm. Nature Rev. Genet. 9, 663–677 (2008).
98. Denes, A. S. et al. Molecular architecture of annelid nerve cord supports
common origin of nervous system centralization in bilateria. Cell 129, 277–288
(2007).
99. Röttinger, E. & Martindale, M. Q. Ventralization of an indirect developing
hemichordate by NiCl2 suggests a conserved mechanism of dorso-ventral (D/V)
patterning in Ambulacraria (hemichordates and echinoderms). Dev. Biol. 354,
173–190 (2011).
100. Satoh, N. An aboral-dorsalization hypothesis for chordate origin. Genesis 46,
614–622 (2008).
101. Gerhart, J. Evolution of the organizer and the chordate body plan. Int. J. Dev.
Biol. 45, 133–153 (2001).
102. Lauri, A. et al. Development of the annelid axochord: insights into notochord
evolution. Science 345, 1365–1368 (2014).
103. Cameron, C. B., Garey, J. R. & Swalla, B. J. Evolution of the chordate body plan:
new insights from phylogenetic analyses of deuterostome phyla. Proc. Natl
Acad. Sci. USA 97, 4469–4474 (2000).
This manuscript proposes that the extant enteropneust, rather than
pterobranch, adult body plan may best represent ancestral deuterostome
characters.
104. Erwin, D. H. et al. The Cambrian conundrum: early divergence and later
ecological success in the early history of animals. Science 334, 1091–1097
(2011).
105. Erwin, D. H. & Valentine, J. The Cambrian Explosion: the Construction of Animal
Biodiversity (Roberts and Company, 2013).
106. Peterson, K. J. & Butterfield, N. J. Origin of the Eumetazoa: testing ecological
predictions of molecular clocks against the Proterozoic fossil record. Proc. Natl
Acad. Sci. USA 102, 9547–9552 (2005).
107. Lacalli, T. C. The emergence of the chordate body plan: some puzzles and
problems. Acta Zool. 91, 4–10 (2010).
108. Garstang, W. The morphology of the Tunicata. Q. J. Microsc. Sci. 72, 51–187
(1928).
109. Lacalli, T. C. Protochordate body plan and the evolutionary role of larvae: old
controversies resolved? Can. J. Zool. 83, 216–224 (2005).
110. Sea Urchin Genome Sequencing Consortium. The genome of the sea urchin
Strongylocentrotus purpuratus. Science 314, 941–952 (2006).
111. Freeman, R. M. Jr et al. cDNA sequences for transcription factors and signaling
proteins of the hemichordate Saccoglossus kowalevskii: efficacy of the expressed
sequence tag (EST) approach for evolutionary and developmental studies of a
new organism. Biol. Bull. 214, 284–302 (2008).
112. Baughman, K. W. et al. Genomic organization of Hox and ParaHox clusters in
the echinoderm, Acanthaster planci. Genesis 52, 952–958 (2014).
113. Delsuc, F., Brinkmann, H., Chourrout, D. & Philippe, H. Tunicates and not
cephalochordates are the closest living relatives of vertebrates. Nature 439,
965–968 (2006).
114. Nielsen, C. Origin of the chordate central nervous system — and the origin of
chordates. Dev. Genes Evol. 209, 198–205 (1999).
115. Gudo, M. & Syed, T. 100 years of Deuterostomia (Grobben, 1908): Cladogenetic
and anagenetic relations within the notoneuralia domain. http://arxiv.org/
abs/0811.2189 (2008).
116. Jefferies, R. P. S. The Ancestry of the Vertebrates (Cambridge Univ. Press, 1986).
117. Swalla, B. J. Building divergent body plans with similar genetic pathways.
Heredity 97, 235–243 (2006).
118. Bone, Q. The central nervous system in amphioxus. J. Comp. Neurol. 115,
27–51 (1960).
119. Wicht, H. & Lacalli, T. C. The nervous system of amphioxus: structure,
development, and evolutionary significance. Can. J. Zool. 83, 122–150 (2005).
120. Cannon, J. T. et al. Phylogenomic resolution of the hemichordate and
echinoderm clade. Current Biol. 24, 2827–2832 (2014).
121. Cameron, C. B. Particle retention and flow in the pharynx of the enteropneust
worm Harrimania planktophilus: the filter-feeding pharynx may have evolved
before the chordates. Biol. Bull. 202, 192–200 (2002).
122. Telford, M. J. et al. Phylogenomic analysis of echinoderm class relationships
supports Asterozoa. Proc. Biol. Sci. 281, 20140479 (2014).
Acknowledgments We thank K. Bertsche (http://wanderingfalcon.com), for the
scientific illustrations, C. Patton and J. Watanabe for photography of invertebrate
micrographs, K. Halanych and J. Cannon for providing pterobranch images, and
J. Fritzenwanker for the German translation of Grobben, and helpful discussions.
We apologize to authors whose work we were unable to cite due to space
limitations, and thank M. Kirschner, A. Pani, T. Lacalli, N. Satoh and N. Holland
for discussions that helped formulate these ideas. Support for this work was
awarded to C.J.L from NASA (NNX13AI68G) and NSF (1258169), to D.M.M. from
NSF (IOS1257040). D.S.R. is supported by the Okinawa Institute of Science and
Technology and the US National Institutes of Heath through grant R01 GM086321.
Work at the Joint Genome Institute is supported by the Office of Science of the US
Department of Energy under contract number DE-AC02-05CH11231.
Author Information Reprints and permissions information is available at
www.nature.com/reprints. The authors declare no competing financial
interests. Readers are welcome to comment on the online version of this paper
at go.nature.com/wmiwjq. Correspondence should be addressed to C.J.L.
(clowe@stanford.edu).
2 3 A P R I L 2 0 1 5 | VO L 5 2 0 | NAT U R E | 4 6 5